Biometric identification

Abstract

A method of biometric identification of a person comprises, during an enrolment phase determining at least one enrolment transfer function for bone conduction of an audio signal through an enrolling user's head. The enrolment transfer function is stored. During a verification phase, at least one verification transfer function for bone conduction of an audio signal through the person's head is determined. It is then determined from the at least one verification transfer function and the at least one enrolment transfer function whether the person is the enrolling user.

The invention claimed is: 
     
       1. A method of biometric identification of a person comprising, during an enrolment phase:
 determining at least one enrolment transfer function for bone conduction of an audio signal through an enrolling user's head; and 
 storing the at least one enrolment transfer function, and 
 during a verification phase: 
 determining at least one verification transfer function for bone conduction of an audio signal through the person's head; and 
 determining from the at least one verification transfer function and the at least one enrolment transfer function whether the person is the enrolling user; 
 wherein, in the enrolment phase and/or in the verification phase, the step of determining at least one enrolment transfer function or at least one verification transfer function for bone conduction of an audio signal through a person's head comprises, while the person is speaking and while the person is wearing an earpiece comprising a first transducer located on an outward facing part of the earpiece and a second transducer located on a part of the earpiece facing into the person's ear canal: 
 determining a plurality of times when the speech contains a specific acoustic class; 
 generating the first transducer output signal from the first microphone; 
 generating the second transducer output signal from the second microphone; and 
 determining the at least one enrolment transfer function or the at least one verification transfer function for said specific acoustic class based on the relationship between the first transducer output signal and the second transducer output signal at said plurality of times. 
 
     
     
       2. A method according to  claim 1 , wherein the specific acoustic class comprises voiced speech. 
     
     
       3. A method according to  claim 1 , comprising:
 while the person is wearing the earpiece, and 
 while the person is speaking:
 determining respective pluralities of times when the speech contains a plurality of specific acoustic classes; 
 generating the first transducer output signal from the first microphone; 
 generating the second transducer output signal from the second microphone; 
 determining a respective transfer function for each of said specific acoustic classes based on the relationship between the first transducer output signal and the second transducer output signal at said respective pluralities of times. 
 
 
     
     
       4. A method according to  claim 3 , wherein the specific acoustic classes comprise specific voiced phonemes. 
     
     
       5. A method according to  claim 1 , comprising:
 determining the at least one enrolment transfer function or the at least one verification transfer function by means of an adaptive filter adapted using a Least Mean Squares method. 
 
     
     
       6. A method according to  claim 5 , comprising determining the at least one enrolment transfer function or the at least one verification transfer function by determining a transfer function that would be required to transform the first transducer output signal into the second transducer output signal, or vice versa. 
     
     
       7. A method according to  claim 1 , comprising, in the enrolment phase and/or in the verification phase:
 determining whether the first transducer output signal and/or the second transducer output signal meets at least one quality criterion; and 
 determining the at least one enrolment transfer function or the at least one verification transfer function based on the relationship between the first transducer output signal and the second transducer output signal only if the at least one quality criterion is met. 
 
     
     
       8. A method according to  claim 7 , wherein the at least one quality criterion comprises one or more of a signal-to-noise ratio of the respective transducer output signal; and a number of useful samples in the respective transducer output signal. 
     
     
       9. A method according to  claim 1 , further comprising, after the enrolment phase:
 while a person is speaking, determining that the person speaking is the enrolled user; 
 determining at least one enrolment transfer function for bone conduction of an audio signal through the head of the person speaking; and 
 storing a new enrolment transfer function for the enrolled user. 
 
     
     
       10. A system comprising a processor, wherein the processor is configured for performing a method according to  claim 1 . 
     
     
       11. A computer program product, comprising a tangible and/or non-volatile computer readable medium, comprising computer readable instructions for causing a processor to perform a method according to  claim 1 .